Data Papers on Papua New Guinea Languages, Vol. 40: Phonologies of Austronesian Languages, No. 2

2 Outline of Phonemes

Chart 1 shows the 17 consonant phonemes and 5 vowel phonemes in Misiman.
Chart 1 : Misiman Phonemes

Consonants

Vowels

p

P"

t

k

i

u

b

b-

d

g

e

P

s

a

m

m"

n
[nql

1

[a K .-,1

w

1

J

Uil

h

• This paper is based on research carried out belwei:n Febniary 1978 and April 15)80. Wc are grateful to nuuiy
Misiman friends Tor lime Ihey have given in helping us leani Iheir language, in particular those I'ruin Siagara Sikat village
where we have liccn living. Our neighbours Sikal Hale and SiiHHly Sikat have answered innumerable questions, llic
following have accompanied u< to Ukaiumpa and lieen good language teachers: Weiya Sabl>uth (Loaga), Napitolai
Mesitan (Siagara), Hedidi Mesake (Boiuu) and Jak Oaunedi (Boiou).

Phonologies of Austronesian Languages, No. 2: 1-24

2 Misinian Phonology Bill Callister

3 Inierprclation

3. 1 The Sla(us ofSefiments which may be Consonant or Vowel

The nonsy lliibic high vocoids [j] and [\j\ are interpreted as/j/ and /w/ respectively when
lliey nil the onset slot in the syllable as shown in (1).

1) /jalijiija/ [JAlijajAl 'happiness'
/wakai/ l\)A'kai] 'breadfruit'

/liwai)/ rUUAnI 'story'

An alternative analysis would be to have hi and IvJ realised as the nonsyllabic high vocoids
(or semivowels U| and |w|) when they All the onset slot in the syllable as shown in (2).

2)

/ia.li ia.ia/

liAlijiUAl

'happiness'

/ua.ka.i/

lyA'kai]

'breadfruit'

/li.uan/

rUUAn)

'story'

An argunicnl in favour of this analysis is that it reduces the number of phonemes posited in
Misiman by deleting the semivowels /j/ and /w/ as distinct phonemes. I have chosen the first
analysis, however, since /j/ is said by some speakers with a definite fricative quality which
would lend to support its status as consonant in its own right. For the remainder of this paper
I will use /j/ and Av/ in both phonemic and phonetic representations.

.V 2 rhc Status of Units which may be Sequence or Unit

.1 2. 1 l.abialised stops. Labialisation is a feature of Milne Bay languages and labialised
segments have regularly been interpreted as units in these languages. Following this
practice, I ain interpreting the sequences |p"l, [b"], [m*] as single units /p"/, /b'/, /m"/.

I'hey cannot be interpreted as sequences of two consonants because there arc no
nonsuspicious consonant clusters except across syllable boundaries. An alternative analysis
would be that the sequences \p"V\, [b'V] and [m"V] are the phonetic realisations of
/pu V/, /buV/and /nui V/, respectively. There are several arguments in favour of this analysis.
First, it reduces the number of phonemes posited in Misiman by deleting /p"/, /b"/ and /m"/
as distinct phonemes. Second, there are no occurrences of /puV/, ^uV/ and /muV/ unless
the labialised consonants arc derived from these sequences. Third, there are no clear-cut
occurrcncesof the vowel sequences Aic/or/uo/. Ifthe labialised consonants were interpreted
as sequence of CV this would introduce at least ten occurrences of /ue/ and two occunences
of /uo/, as illustrated in (3).

Phonologies of Austronesian Languages, No. 2 3

3) a. /mu'egan/ ('ni'^an] 'a sail'

b. /muck/ [m'sk] 'kind of fruit'

The two arguments against this analysis arc botii related to morphology and
n)orphophonemics. First, when a verb is negated with the morpheme /nige/ 'not', the first
syllable of a verb stem is reduplicated as shown in (4).

4) a. /pek/ [pek] 'give'

/ni'ge ipe'pek/ [ni'ge ipe'pek] 'he did not give'

b. /ba'teli/ [ba'teli] 'promise'

/ni'ge ibaba'teli/ [ni'ge ibaba'teli] 'he did not promise'

When the first syllable of a verb stem starting with a labialised consonant is reduplicated,
the reduplicated syllable consists of the labialised consonant with the following vowel as
shown in (5).

5) a. /p^ela/ ['p*ela] 'open'

/ni'ge ipVp^ela/ [ni'ge ip*f'p''ela] 'he did not open'
b. /b'ata/ [■b"'ala[ 'be big'

/ni'ge ib'a'b'ata/ [ni'ge ib^afa'ata] 'it is not big'

This reduplicated syllable cannot be interpreted as/Cu V/ (viz. /puc/ in our example) because
there are no examples of non-suspicious syllables which when reduplicated show a complex
nucleus of two vowels. It must be interpreted as [C'V] (viz. /p"c/ in our example).

The second argument against the analysis of the labialised sequences as /CuV/ is related
to morphophonemic changes discussed in 5.9 involving /p/ and in 5.7 involving /m/ which
are more readily understood in terms of a feature oflabialisalion which is added in certain
contexts than in terms of a IvJ inserted after the /p/ or Iml which is in turn realised as [p"*)
or |nr|. The causative prefix {pa-} has the two allomorphs [pa] and [p'al; while final /m/
on word stems is labialised when followed by a vowel due to reduplication or addition of a
suffix. These morphophonemic changes may be easier to explain in terms of an added feature
of labialisation may in terms of a morphophonemic change from /p/ to /pu/.

3.2.2 The Interpretation of Vocoid Sequences. ' The ambiguous vocoid sequences [ai[. [au[,
[eil. [cu|, |ia|, [ic], [io|, (iu|, [oi|, [ou|, [ua|, [ui| have not been interpreted as units for several
reasons. First, a number of unambiguous sequences occur as shown in (6).

6) a. [ae] /ba'ewa/ [ba'ewa] 'shark'
b. |ao] /ikao'me/ [ikaa'me] 'he stole it'

' Sc(|uci)ccs of itlcittical vitc-ujib do iiiH occur CNccpt iicniit.s luoiplicinc iKuimljiricit.

lisin

iiaa Phonology

Bill Callister

c.

[(•a) /Bi'neja/

(Dinea) 'net'

d

1 ,)<(1 /imo'wasi/

[ini^'wasi] 'it is Tinished'

c.

|>tj /bo'elii/

[b^'tiul 'kind of grass skirt'

Second, sets of icversed sequences occur including [ai] and [ia], [au] and [ua], [ei] and [ie],
|io| and |oi|, and [iu| and [ui]. fhird, the second vowel of the sequence is often stressed,
showing them (o be nuclei of two separate syllables as seen in (7).

7) a. /ka'un/ [ka'un] 'head'

b. /ko'inian/ [ka'iman] 'kind of ant'

Fourth, given a stem-final vocoid sequence which ends in /i/ or /u/, the I'll or IwJ changes to
/j/or/w/ when a vowel-initial suffix is added. The other vowel, however, remains unchanged.
Examples are given in 5.6,

Thus, all vocoid sequences are interpreted as sequences of phonemes. The sequence of
any vocoid followed by either a higher vocoid or a vocoid of the same level is interpreted as
VV The sequence of any vocoid followed by a lower vocoid, however, is interpreted as VCV
where C is transitional /w/ or transitional /j/, as shown in (8).^

8) a. li.i'l is /ije/ as in /bagijewa/ 'coconut'

b. |i }1 is /ijo/ as in /ahamisijo/ 'chair'

c. |i.a) is /ija/ as in /gijal/ 'dig'

d. |c a] is /cja/ as in /Bineja/ 'net'

e. |u (■] is Aiwe/ as in /luiwe/ 'here'
f |ii .7I is /iiwo/ as in /huwo/ 'there'
g. |u.a| is Aiwa/ as in /huwal/ 'comb'

h. I^.al is /owa/ as in /lowakeb/ 'squabble'

An allcrnaii\ c analysis here would be that the sequence /V, V^/ is realised phonetically
as I Vi'VJ or I V|'VJ when V, is a higher vocoid than V^, that is /ie/ and /ua/ are realised
as |i'c| and |u"a|, respectively.

There are a number of arguments in favour of this analysis. First, using this rule it is
possible to keep an underlying form of /W/ but still account for the presence of /w/ and
/j/ in spoken speech. Second, the same rule applies both within and across morpheme
bounda ries. Third, reduplication patterns seem to indicate sequencesofV,.Vj are phonemically

'The cKamplcb lisle J for sci{ucnces |ut:| and |uo| arethe only occurrences of these sequences. However both ttiese
woids have alicniaiivc tonus /hiwc' 'here* and j^iiwo/ 'tliere'. Tlie /w/ caiuiot lie regarded as transitional /w/ in these
lonns, hut ab there in its own right, i lence it is reasonable to believe that similarly it is not a transitional /w/ in the tbnns
. lm\se/ and yhiivvx/. It this is so, there are no examples of the sequences |uf| or |u.7).

The tollovviiig woi 1I.S have /j/ where it would not Ire piedieted from lite above rules: /ejowa/ 'garderr'. /Iieju/ 'put
siieks across a gaiden', ''l»"eju/ 'previously cooked food'. This /j/ is interpreted not as a transitional /j/lwtween two
vowels but as Ireuig there in its own rigllt.

Phonologies of Austronesian Languages, No. 2 5

A'V/. Wlien adjectives are formed from verbs, the first syllable of tlie verb root is normally
reduplicated. However when the verb root begins with the sequence CW the first two
syllables (that is both vowels) are reduplicated. For example, /waisi/ 'be good' forms the
adjective /waiwaisana/ 'good'. Sequences which involve a transitional [j] or [w] also behave
in the same way. For example /mo'asi/ [m^'wAji] forms the adjective /moamoa'sena/
[m9WAm7WA'jenA] 'last (one)'. Fourth, it is possible to outline factors affecting the degree
of realisation of the transitional /j/ or /w/. These differences include the difference in height
between V^ and V,, and stress. The transitional semivowel is more pronounced the greater
the difference in height of V, and Vj, or if either V, or Vj is stressed. For example, the [w]
is more pronounced in /mo'asi/ [m^'wAji] 'be finished' than in /moamoa'sena/
(ni^'WAni.jwA'jenA] 'last (one)'.

There are several arguments against this analysis, in favourofincludingthe semivowels
in the phonemic forms. First, psycholinguistic testing shows that native Misiman speakers
are very conscious of the transitional /w/ or /y/ as being present in the above sequences. All
sequences have been tested informally. A formal test was also conducted which concentrated
on vowel sequences which were difTicult to interpret. In the printed literature, a hymn book
and old translation of the New Testament, vocoid sequences have usually been written
without an intermediary <y> or <w>. Despite this fact the test showed a strong preference
to insert <y> or <w> in sequences of a vocoid followed by a lower vocoid. In those words
in which a medial <y> is possible there was an average result of greater than 13 out of 15
preferring the form with <y> present. In those words in which a medial <w> is possible there
was an average result of greater than 14'/j out of 15 preferring the form with <w> present.

Second, some speakers give a fricative quality to the transitional /j/. For example,
/nigeja/ not' varies between [ni'gcja] and [ni'geja]. Third, the transitional /j/ and /w/ are
inserted across morpheme boundaries as shown in (9). (See 5.5 for further discussion.)

9) a. /ijab/ 'she gave birth' from /ab/ 'give birth'

b. /uwegon/ 'you (sg.) went' from /egon/ 'go'

c. /sab"eluwa/ in the sun' from /sab'elu/ 'sun'

4 Description of Phonemes

■/. / Seiiinenlol Items

The phonemes /tds/ are manifested as the dental phones [}^s] (although /d/ is not very
coninion); /n/ is realised as the velar (i]| Ixifore a velar stop, and as the dental (n| elsewhere.
The lateral approximant /I/ is realised as the retroflexed [IJ. The front semivowel /j/ may have

6 Misiinaii Phonology Bill Callister

the fricative quality |j | Tor some speakers. Examples illustrating each consonantal phoneme
arc given in Apixjndix A.

The nnal consonant ofcertain words is occasionally given nasal release when the word
is said with emphasis. It is not a very common phenomenon, and only occurs utterance final
and when ihc final syllable of the word is stressed.

10) a. /I'nak/ li'ijak'] it is no good.'
b. /jawa'tui/ |ja\va,(uj'] i'mcold.'

The mid vowels /e o/ are manifested as the mid open vocoids [e 5]. In one word, [m'3k]
'kind of fruit', /o/ is nasalised; there is no other instance of nasalisation in Misima.

The low vowel /a/ is, as a norm, manifested as the low close unrounded vocoid [a].
However it ranges from the mid open unrounded vocoid [o] to thelow open unrounded vocoid
[a] The factors which lend to make the vowel lower are stress, the environment of the labials
/I)/ and /w/, and 1 he environment of the velars /k/ and /g/. When stress is combined with one
of the segmental environments, the phoneme /a/ reaches the lowest level of the continuum,
the allophonc |a|, as shown in (1 1).

11) a. /lonowak/ [(9?)?'wak] 'man'

b. /(iaOa/ [BafiA] 'belong (of hair)'

c. /gaga/ I'gagA] 'lice'

The factors which tend to make the vowel higher are lack of stress; the environment of the
alveolar continuants /s/, /n/, and /I/; and the environment of /m/. When the lack of stress is
combined with one of (he segmental environments, the phoneme /a/ reaches the highest level
of the continuum, allophone la], as shown in (12).

12) a.

/lalam/

[.(Alom]

'allow'

b.

/salija/

[53'liJA]

'blood'

c.

/liinan/

I'liinaol

'mother'

In addition 10 the five vowels described thus far, a contrast has been found between [o]
and [?| in a few words.

13) a. |o|| 'crushed to powder'
b. I.7II 'chew betel-nut'

14) a. |'!i)ii'|on| 'group of three coconuts'
b. |l?n'J?i,!| 'burning'

Sinularly a coiiiiasl has been found between /e/ and /d in a few words.

15) a. It'll 'part of a boat'

b. |i II 'cut oul inside of

Phonologies of Austronesian Languages, No. 2 7

16) a. Ilewa] 'be aflected by betel-nut'
b. l[ewa] 'kind of tree'

Native speakers find it very difllcuit to distinguish between /o/ and hj and between
Id and kl except in the case of the few minimal pairs. For example, when asked if the e in
two different words is the same or different, they give varying answers, even after they have
been made aware of the difference in the minimal pairs.

The phones [e] and [o] cannot be interpreted as the sequences /ei/ and /ou/ since these
sequences occur in other forms as shown in ( 17).

17) a. /keke'isi/ [keke'iji] 'little'

b. /toun/ ['|?uq] 'carry on head'

They also cannot be interpreted as the sequenccs/ee/ and/oo/ since I here arc no unambiguous
sequences of identical vowels except across morpheme boundaries. These two phones
remain as unsolved problem.

■f 2 Stress

Stress is phoiiologically unpredictable and is therefore phonemic. It is marked by a
combination of intensity, length and pitch. The stressed syllable is usually lengthened,
especially when it is the first syllable of a two syllable word. The lengthening of the syllable
is luiniiual when the stressed syllable is word final. Stress is also marked by greater intensity
and higher pilch on the stressed syllable.

Many words also have a definite secondary stress which iKCurs on ihc second syllabic
preceding the syllable with primar>' stress, and to a lesser e.xlcni on alternate syllables
preceding this one.

■f 2.1 Stress on nouns. Most two-syllable nonreduplicated nouns arc stressed on the first
sy 1 lable as shown in ( 1 8a, 1 9a), but some are stressed on the second sy 1 lable as shown i n ( 1 8b,
19b).

18) a.

/b'-alal/

[■b"a:laU

'floor'

b.

^"a'la/

[b-a'la]

'side of boat'

19) a.

/hiwo/

[hi.w?!

'fire'

b.

/hi'wo/

[hi'w?!

there'

Two-syllable reduplicated nouns have stress on both syllables as can be seen by comparing
Ihc reduplicated form in (2()a) with the nonreduplicated form in (2()b)

8 Misiman Phonology Bill Callister

20) a. /aiVan/ I'ao'at)] 'food' (from /an/ 'eat')
b. /anan/ ['aijai)! 'voice'

Wlicii (wo-syllabic reduplicated words are used in a flow of speech, the second syllable
receives greater stress when the word occurs utterance initially as shown in (21a); the first
syllabic receives greater stress when the word occurs noninitially as shown in (21b).

21) a. /kam'kamja'kite/ [.kam'kamja'kije] 'I sawthehen.'
b. /no kani'kam te/ (ijrj 'kaiii,kamje] 'This is my hen.'

Most three-syllable nouns are stressed on the penultimate syllable as shown in (22),
although a small group are stressed on the final syllable as shown in (23).

22) a. /kalehe/ (ka'lehe] 'mango'
b. /ginaha/ (gi'ijaha) 'fire'

23) a. /sina'wel/ [jina'wcU 'creek'

b. /ulabo/ [ula'b?] 'crayfish'

All four-syllable nouns except reduplicated words are stressed on the penultimate
syllable, with a secondary stress on the first syllable.

24) a. /lele'waga/ [idc'waga] 'crocodile'
b. /ata'kena/ |,aja'ktna| 'cockatoo'

Four-syllabic words that are reduplications are stressed onthe firstand third (i.e. penultimate)
syllables.

25) a. /bili'bili/ [bili'bili] 'ground'

b. /kasi'kasi/ I'kasi'kaji] 'mangrove'

There arc no simple nouns of more than four syllables.

When the locative suffix {-a} is added to nouns the stress always falls on this suffix.
If the noun root already ends in /a/ then the stress simply shifts to the final /a/ without any
overt locative suffix as shown in (26).

26) a. /hoga/ ['h?ga] 'sea'

b. /lio'ga/ [hcj'ga] 'at/to the sea'

■4.2.2 Stress on verh.s. Stress on verbs is generally predictable, and is closely tied to the
transitivity of the verb. Whenever a verb root has both a transitive and intransitive form, the
intransitive form is stressed on the penultimate syllable. The transitive form of some two

Phonologies of Austronesian Languages, No. 2 9

syllable verb roots is formed simply by moving the stress to the final syllable as shown in

(27-2K).

27) a. /i'gijal/

[i'g'JaU

'he dug'

b. /igi'jal/

(igi'JaU

'he dug it'

28) a. /i'fialuk/

[i'Oaluk]

'he planted'

b. /ifialuk/

[iOaluk]

'he planted it'

The transitive forms of other verb roots involves a vowel change of the final vowel or the
addition of a suffix, but in these cases too the stress is moved to the final syllable as shown
in (29).

29) a. /i'pani/ [i'pa^ii] 'he climbed'
b. /ipa'ne/ [ipa'iie] 'he climbed it'

Almost all intransitive verbs without corresponding transitive forms are stressed on the
penultimate syllable.

30) a. /itu'keli/ [i^u'kelil 'he coughed'
b. /i'niCal/ [i'niUalJ 'he laughed'

Two exceptions are /lo'm'an/ 'be hurting' and /hou'wa/ 'go first'.'

Most transitive roots without a corresponding intransitive root are stressed on the final
syllabic.

31) a. /iki'te/ [iki'je] 'he saw it'

b. /igi'nol/ [igi'ngU 'he made it'

c. /itou'lil/ li,l3u'lilJ 'he followed him'

However there arc several examples ofa transitive root stressed 011 ihc pcnultiniale syllable.

32) a. /i'kasi/ [i'kaji] 'he scraped it'

b. /i'lJihi/ [i'liihi] 'he cut it out (of canoe)'

There is at least one pair of transitive verb roots which contrast in stress as shown in (33).

33) a. /i'gabom/ [i'gabom] 'he shared it out'
b. /iga'bom/ [iga'bam] 'he broke it'

When a verbal prefix is added to a two-syllable verb root with stress on the final syllable,
there is definite secondary stress on the antepenultimate syllable.

34) /nihila'be/ [ni.hila'be] 'they will help him'

' III tlie case (>r/1iou'wa/ tli>: stress may Iv on (he final syllable due tu a locative sutri.N (a), discussed below.

10 Misiman Phonology Bill Callister

Most verbs may take locative suffixes which attract stress. One of the suffixes is {-a}.
If this siitfix is added (o a verb root ending in /a/ then the stress simply shifts to the final
/a/ without any oven locative suffix as shown in (35).

35) a. Auiawa/ [uijawa] 'you fished with a spear'

b. /(ga) uila'wa/ [(ga) ui|a'wa] '(where) did you fish?'

Most verb stems can be reduplicated. When a stem is partially reduplicated there is still
only one primary stress. However, when verb stems of one or two syllables are reduplicated
fully both parts of the new stem receive primary stress.

.36) a. /i liblib/ [iiibiib| 'he keeps on grabbing'

b. /i'logi'logi/ (i'l?gi'l?gil 'he is fishing with a net'

Some verb stems are always reduplicated. These also have two stresses if they are fiiUy
reduplicated.

37) a. /i'anan/ [i'aij'ao] 'he ate a meal'

b. /ikeUakeOa/ [ikcBakcBa] 'it is dry'

4 2.3. Other .SiiL'.'i.\ ( ienerali.salion.y. Compounds are stressed according to the roots that
iii;ikc ilicni up, with the primary stress on the final stem and secondary stress on the other
root or roots of the compound.

38) a. /aba#lap*a'lolo/ (,abaiap'"a'[9l3l 'church' (lit. 'place of worship')
b. /ali#lo'm*an/ |,atil3'm"an | 'pity' (lit. 'heart-pain')

Most grammatical words are stressed on the penultimate syllable as shown in (39).

39) a. /cbo/ (cbD] 'if

b. /e'hila/ Ic'hila] 'how many'

Grammatical words stressed on the final syllable include exclamations as in (40a), locative
words as in (4()b), some numerals as in (40c), /a'ge/ 'question word', and /a'ma/ 'we (excl)'.

40) a. /a'hi/ (ahi] '1 don't know'

b. /ja'ho/ Ija'h?] 'over there'

c. /c'pal/ (L-'paJl 'four'

One of the locatives with final stress contrasts with a noun with penultimate stress as shown
in (41).

Phonologies of Austronesian Languages, No. 2 11

41) a. /hiwo/ [hiwa] 'fire'

b. /hiwo/ [hi'w?] 'over there'

Wl\en a word which is usually stressed on the final syllable is utterance final, the
primary stress is fiequently moved to the antepenultimate syllable. For example, when the
transitive verb /jaki'ie/ 'see' is nonfinal the stress is final as in (42a), but when it is utterance
final the stress is antepenultimate as in (42b).

42) a. /ga'lokjaki'tebu'bun/ Iga'lakjaki'Je bu'bun] ' I saw the girl clearly'
b. /ga'lok jakite/ Iga'lnk'jakije] '1 saw the girl'

Similarly, while stress usually occurs on final syllable with the locative suffix as shown in
(4.1a), it shifts to the antepenultimate syllable when it is utterance final as in (43b).

43) a. /ikite'wa b'agaoia/ [ikite'wa b'agaoia] 'HesawyouinBwagoia.'
b. /ga i'kitewa/ [ga i'kitewa] 'Where did he see you?'

■/. 3 Intonation

Contrastivc intonation patterns have been found between statements, questions, and
pause. Statement intonation is marked by a slight rise in the penultimate syllable falling to
a low on the final syllabic as in (44).

44) /igaga'jawa/ ligaga'jawa] "he was watching'

Question intonation is markcdby a high pilch on the penultimate syllable, which is sustained
on ihc final svllable.

45) /igaga'jawa/ [igaga'jawa] 'Was he watching'.''

Pause intonation is marked by a rising pitch on the final syllable.

46) /igagajawa/ (igaga'jawa] 'he was watching...'

5 Morphophonemics

5. / Changes in Vowel in the Word Stem

Alternations occur between the high vowels /i u/ and the mid vowels /e o/ in several
environments. Whenan intransitive verb stem ends in a vowel the most common way to form
the transitive stem involves this lowering as shown in (47).

12 Misimaii Phonology Bill Callister

47) a /ipaiii/ he climbs' /ipa'ne/ 'he climbs it'

b. /bcjii/ 'put sticks across' /be'jo/ 'put stick across it'

The second environment involves possession. There are three types of possession in
Misima Inalienable possession is Tormcd usinga siifTix. For third person singular, this suffix
is { -na > When ( he free form of Che noun ends in a high vowel it frequently lowers in the third
singular form as shown in (48).

4K) a /c'loki/ 'siomach' /elo'ke-na/ 'his stomach'

b /c'noiui/ 'oil' /eno'no-na/ 'its oil'

Finally, udjcciivcs arc formed from verb stems by reduplicating the first syllable of the
verb stem and addi iig an agreement suffix. For the third person singular theagreement suffix
is /-aa/. When Ihc verb stem ends in a high vowel it frequently lowers in this form as shown
in (49).

49) Aiiki/ 'be black' /bibi'kena/ 'black'

One common word, /wa'isi/, is exceptional in that the final I'll becomes /a/ instead of /e/ as
shown in (50)

50) Ava'isi/ be g(X)d' /waiwai'sana/ 'good'

When ihe final vowel is /a/ in one of the environments, it also changes, although in
difft rent ways. General ly, it becomes Id i n forming the transitive verb stem as shown in (5 1 ).

51) /i'honii/ "he pushes' /iho'ne/ 'he pushes it'

If, however, the verb takes overt transitive suffixes for persons other than third singular, it
is not changed as shown in (52).

52) /ip'awii/ 'he found (it)' (cf /i'p-'awa-l/ 'he found them')

This contrasts with similar verbs ending in l\l and Iwl in which the vowel does change in
forming (he iransifive stem as shown in (53).

53) /ii'eli/ 'he put (it)' (cf /itele-I/ 'he put them')

Stem final IaI generally remains unchanged in forming inalienable noun stems and
adjective stems as shown in (54-55).

54) /salija/ blood' /sali'ja na/ 'his blood'

55) /b'ata/ 'be big' /b''ab"'a'ta na/ 'big'

Thercarca few exceptional stems which form inalienably possesscdfornis by changing
final IaI to Id as in (56) or by adding an Id as in (57).

Phonologies of Austronesian Languages, No. 2 13

56) /"mwaha/ 'dish' /mwa'hena/ 'his dish (of food)'

57) IVjJ 'excreta' Aa'ena/ 'his excreta'

5.2 Transitional /i/

When a stem ends in a consonant, the transitional vowel /i/ is inserted before any siifTix
beginning with a consonant. This occurs in inalienably possessed nounsas in (58), adjectives
as in (59), transitive verbs as in (60), and directional affixes as in (61).

58) /hul/ 'breast' /hu'lina/ 'her breast'

59) /ket/ 'be red' /keke'tina/ 'red'

60) /lol/ 'to hit" /ilo'lil/ 'he hit them'

61) /ip-'ata'n-ik/ 'he look it to there'

5.3 Deletion of /n/

The most frequent pattern for forming the inalienably fwssessed noun stem is to delete
final /n/ on the free form of the noun.

62) /tanan/ 'ear' /ta'na-na/ 'his ear'

If, however, the free form ends in /un/ the stem is formed by adding /u/.

63) /ka'un/ 'head' /kau'nu-na/ 'his head'

5.4 Vowel Elision

The suffixes listed in (64) begin with vowels. In each case the suffix attracts stress.

64) a. /-'au/ plural suffix on personal nouns

b. /-'au/ first singular short transitive suffix

c. /-'a/ locative suffix on nouns, verbs and some locatives

Ol her vowel initial suffixes include a set of long transitive suffixes of which the form for third
person singular is/-'an/ or /-'i/, and the marked directional suffixesof which the formfor third
person is /-'ek/.

14 Misinuin Phonology Bill Cailister

When a sufTix beginning in/a/is added tea stem ending in/a/, one /a/ is generally elided
as sliown in (65).

65) a. {wawaja-'au} /wawajau/ 'children'

b. {igagajawa-an} /igagaja'waiv/ 'he watched it'

c. {hoga-'a> /lioga/ 'at the sea'

However final /a/ on a verb stem need not be elided, especially in slow speech, before any
of the long transitive suffixes other than the third person singular. In these cases the stress
does not fall on the /a/ of the suffix. Compare (66) with (6Sb).

66) ( igagayawa-a'gil } /igagayawaa'gil/ 'he watched them'

When the directional suffix /-'ek/ is added to stems ending in a vowel, the stem final
vowel generally is deleted.

67) a. {jakite-'ek} /jaki'tek/ 'I saw it in that place'

b. {imina-'ek} /imi'nek/ 'it stays in that place'

c. {ijalijoho-'ck} /ijalijo'hek/ 'he threw it away over there'

In one common word, stem final lol may be retained before /-'ek/ as shown in (68).

68) {ino-'ck} /ino'ek/ 'he went by that way'

There arc no examples of elision occurring when a prefix ending in a vowel is addcil
10 a sicni beginning with a vowel. Instead, both vowels remain as shown in (69).

69) a. {pa-an} /pa'aii/ 'food' ('CAUS' + 'eal')

b. {i-'inni} /i'imii/ 'it is sharp' ('.Is' + 'sharp')

c. {e-'esoni} /e'esom/ '(coconut) husker' ('INST' + 'husk')

y 5 Transitional /]/ and /w/ across morpheme boundaries

Whenever the addition of either a prefix or a suffix to a stem results in a vowel sequence
ill which the first vouci is higher than second, a transitional /j/ or /w/ is inserted between
the two vowels agreeing in backness with the first vowel. Transitional /j/ and /w/ sound the
same as /j/ and /w/ that are part of the stem.

70) a.

{i-'ab>

/ijab/

'she gave birth'

b.

{i-japu)

/i'japu/

'it is long'

71) a.

{ku-ol}

/ku'wol/

'you (pi.) chew betel nut

b

{kii-'woli}

/ku'woli/

'you (pi.) complain'

PlioiiologiesofAustronesian Languages, No. 2 15

5 6 OesylUihiJlcation of A/ antl/u/

Given a slem-final vocoid sequence which ends in /i/ or /u/, the /i/ or /u/ changes to
/j/ or /w/ wl\cn a vowel-initial suffix is added.

72) a. /ikali'liai/ 'he carried (it)'
b. /ikaliCa'jau/ 'he carried me'

73) a. /pwa'siu/ 'platform'

b. /pwasi'wii/ "on the platform'

5 7 Lahialisalion of An/

A sicin final /ni/ is labialised whenever it is followed by a vowel resulting from the
addition of a sufTi.x or from reduplication.

74) a.

/heli'jam/

'friend'

b.

/helija'nrau/

'friends'

75) a.

/pwalum/

"seed'

b.

/p\valu'm*ina/

'its seed'

76) a.

/im/

'drink'

b.

/i'iu*im/

'get drunk

I here arc two nouns in which the inalienably possessed stem ends in hnaJ. In the first and
second singular possessive forms, this /m/ is also labialised even though il is not stem final.

77) a. {'niman} /niman/ 'hand'

b. {'niman-na} /ni'mana/ 'his hand'

c. {'niman-u} /ni'm^au/ 'your hand'

y S Lahialisotion of/p/ and /h/

In contrast with stem-final Iml. stem-final /p/ and ^/are not generally labialised when
followed by a vowel as shown in (78).

78) a. /pap/ 'gum' /pa'pina/ 'its gum'

b. /ab/ 'give birth' /ab'ab/ 'give birth repeatedly'

There are, however, a few examples of stem final /p/ and /b/ being labialised when a sufilx
is added as shown in (79).

16 Misiniaii Phonology Bill Callister

79) a /opop/ a depression' /opop'ina/ 'its depression'
b. /aliiwab/ north' /aluwa'b'a/ 'on the north (side)'

The lack of enough examples makes it impossible to formulate a rule.

5. 9 'I he aiiisdiive ami slalivc prefixes

The causative prefix {pa-} has the allomorphs /pa-/ and /p"a-/; the stative prefix
{ma-} lias ilic allomorphs /ma-/ and /nra-/. The allomorphs with labialised consonants
general ly occur whenever the stem begins with a labialised consonant or with the sequences
/nui/, /mo/, /bii/ or /bo/ as shown in (8()a,b) and (8 1 a,b). The stative allomorph /nra-/ occurs
more generally whenever the stem begins with the sequence /Cu-/, /Co-/ or /Ca-/ as shown
in (81c).

80) a. /p'a'nral/ 'save'
b. /p*a'mola/ 'buy'

81) a. /nra'gabom/ 'be broken'

b. /nralonra/ 'be cut'

c. /m"ahawan/ 'have a hole in it'

Some roots, however, may take either /pa-/ or /p'a-/. Some speakers, especially younger
people, show a preference for the /pa-/ allomorph.

82) a. /p'anre'jaha/ - /panrejaha/ 'do quickly'
b /p'a'iimsa'Muisa/ - /panuisamusa/ 'make dirty'

5. 10 F.lisUm of the vowel following /h/

A number of \ crb stems end with the sequence /V^hV/. For five of these verbs the stem
is reduplicated in the intransitive form.' In these forms the final vowel elides and the stress
shifts as shown in (81).

S.T) a. /ni'bilii/ 'smack' /i'nibi'nibih/ 'he is smacking''

b. /ja'muhu/ 'wash sago (trans)' /i'jamu'jamuh/ 'he is washing sago'

* Twu vi;rli!i orthi:> font) i)o nol have a corresponding reduplicated form for semantic reason, viz. /alalia/ 'deny*
and /l>a'kuhu/ 'cease'.

' Some Misiiiians have told me that there is nn/V on the endof /i'nihi'nihih/ 'he is smacking'. If this is so the/h/
has tieeii dropped as uell.

Plioiiologies of Austronesian Languages, No. 2 17

One of tliesc verbs, /ja'muhu/ "wash sago' also has an unredtiplicated intransilive form in
which the final vowel is dropped,

K4) /i'jamuli/ 'he washed sago (inlrans)'

All syllable-final occurrences of /li/ arise through this process.

Two verb stems ending in /V|hV/, /a'waha/ 'weed' and /cBe'ehe/ "steer', do not elide
the final vowel in reduplicated forms,

K.S) /ie'Bec'Ikhe/ 'he is steering'

When a suffix is added to these stems the final vowel is usually elided.

X6) a. /inibihau/ 'he smacked me' (*/inibihe'jau/)

b. /illikuhij;»/ 'he used it to mix with' (*/iliikuhui'j;i/)

There is one stem ending in /V|hV/ in which the final vowel is not deleted before a suffix.
87) /gogehcgehejan/ 'cut lengthwise lots of times' (*/gogehegeh'jan/)

6 DiNtrihution

(>. I Syllnhles

The following syllable types occur: V. CV, CVC, VC. The open syllables V and CV are
found ill all positions of the word. The closed syllables CVC and VC arc found only word
finallv and across morpheme boundaries, with the exception of the fixe words in (SS).

88) a. /amna/

'feelings'

b. /kimpos/

'centre pole'

c. /kin'b*ai/

'fingernail'

d. /sin'gili/

'centre rib of coconut leaf

e. /tab'naha/

'sneeze'

The maximum number of syllables inanonreduplicated root is four asshown in (89a). Words
may be up to nine syllables in length as shown in (89b).

89) a. /ba.gi.'c.wa/ "coconut'

b. /ni. hi. lata. hi. a. gi. mill/ 'they will leave you (pi)'

18 Misiman Phonology Bill Callister

All coiisoiiaiils may fill (he onset slot. The coda slot may be filled by the two nasals
Ixnl and Inl; lsl\ l\l\ and by any nonlabialiscd stop except /d/.' The phoneme /h/ also fills the
coda slot in a few words as noted in 5. 10. The only monomorphemic consonant clusters are
/mn/, /nip/, /nb/, /ng/, /bn/. These always occur across syllable boundaries.

Any vowel n)ay fill the nucleus position. There are no occurrences of sequences of
idcMiical vowels except across morpheme boundaries.

The only rest rici ions on C V sequences are that the phonemes /b* m* w/ are not followed
by Ai/; /B/is not followed by Ai/ or /o/; /p"/ is not followed by /u/, /o/ or /i/; and/j/is not followed
by Ixl.

Almost all poiciuial sequences of two vowels or of two vowels with transitional /j/ or
/w/ occur with the exception of /eo/,' /uwo/ and /uwe/, as well as any sequence of /Vji/ or
/Vwii/ as noied above. The sequences /uwo/ and /uwe/ are discussed in footnote 2. An
alternative analysis which would introduce two of these sequences is presented in 3.2.1.

Data Papers on Papua New Guinea Languages
Volume 40

Phonologies of

Austronesian Languages

No.2

Edited by John M. Clifton

Summer Institute of Linguistics
Ukarumpa via Lae
Papua New Guinea

Papers in the series Data Papers on Papua New Guinea Languages (formerly
Workpapers in Papua New Guinea Languages) express the author's Icnowledge at the
time of writing. They do not necessarily provide a complete treatment of their topic.
However, it is felt that these papers should be made available at this time.

John M. Clifton. Series Editor

Copyright © 1993 by the

Summer Institute of Linguistics

Ukarumpa via Lae

Papua New Guinea

Published 1993

Printed by the SIL Printing Department
Ukarumpa via Lae
Papua New Guinea

ISBN 9980 1095 9